Georgia Institute of Technology
AI-Based Analytics and Energy Modeling Frameworks for Characterizing Urban Energy Systems
Abstract
dc:description.abstractUrban energy systems are growing in complexity as they respond to increasing demands for efficiency, resilience, and decarbonization. However, current modeling approaches often fail to capture the physical, behavioral, and systemic diversity needed to support effective planning and decision-making. This thesis develops an integrated, framework that combines bottom-up physics-based modeling and simulation, behavioral modeling, and AI-driven analytics. Across three chapters, I present a cohesive and adaptable approach for characterizing and analyzing urban energy systems at scale. Chapter 2 establishes the core modeling foundation, enabling district-scale simulation of buildings, grid infrastructure, DERs, and emissions. The framework supports scenario-based analysis and co-design of energy systems. However, physical modeling alone is insufficient, as neglecting how humans interact with these systems introduces bias in both the model’s structure and its outcomes. Chapter 3 addresses this gap by embedding behavioral complexity into the modeling framework. The core framework is extended to capture behavioral complexity through personalized, stochastic, and agent-based modeling. By incorporating human behavior into the modeling structure, the framework captures the diverse of energy use patterns reduces bias in urban energy analyses. Chapter 4 introduces a reverse-engineering framework to address data gaps and scale the urban energy modeling and analysis process. I develop AI-based methods to automate the generation of missing inputs and enable dynamic scenario exploration. These methods are designed to handle the multimodal nature of UBEM data—including geometric, numerical, categorical, and time series inputs. This approach significantly reduces the labor required for model preparation and expands the analytical space available to energy planners. Additionally, it ensures that AI-generated inputs and scenarios remain physically plausible through dual validation—against both test datasets and UBEM simulations. Together, these contributions represent a unified vision: a scalable, more accurate, and localized framework for modeling and optimizing urban energy systems. This work enables decision-makers, planners, policy makers, utilities, and building owners, to explore strategies, behavioral, and system interactions, and generate future scenarios. Looking ahead, these capabilities establish a foundation for intertwined learning, where buildings, DERs, the grid, and market sectors dynamically interact to support complex, real-world decision-making.
Degree
thesis:*- Level thesis:degree_level
- Doctoral
- Department dc:contributor.department
- Architecture
- Grantor dc:publisher
- Georgia Institute of Technology
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- El Kontar, Rawad
- Advisor dc:contributor.advisor
-
- Rakha, Tarek
- Committee members dc:contributor.committeemember
-
- Krietemeyer, Bess
- Taylor, John E.
- Kastner , Patrick
- Chintala, Rohit
Subjects
dc:subject × 10Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/82011